Every intake I get one student who has decided he is an attacking player. He has read a book with the word sacrifice on the cover. He won with the four move mate at 1100 and has spent two years being personally offended that it stopped working. When one of them turns up I put openings away for a month and we play one pawn structure from both sides, twenty minutes a game, no notes, until he stops asking me when the tactic arrives.
The cleanest route is the Queen's Gambit Declined, Exchange Variation: 1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.cxd5 exd5, continuing 5.Bg5 Be7 6.e3 c6. White has traded a c-pawn for Black's e-pawn. That single swap is the entire structural event, and everything for the next thirty moves follows from it.
White: a2, b2, d4, e3, f2, g2, h2.
Black: a7, b7, c6, d5, f7, g7, h7.
Two pawn islands each. Black has three pawns to two on the queenside. White has four to three on the kingside. The c-file is half open for White and the e-file is half open for Black.
The same skeleton appears from the Nimzo-Indian when White takes on d5, and with the colours swapped in the Exchange Caro-Kann and in Scandinavian lines where it is Black holding the two pawn side. It is named for the 1923 tournament at Carlsbad, now Karlovy Vary, where it kept turning up.
Nobody has a diagram in front of them, so here it is in words. White has castled short. Bishop on d3, pointing at h7. Queen on c2 behind it. Bishop on g5, leaning on the f6 knight. Knight on c3. The other knight on e2 rather than f3, which surprises students every single time, because it keeps f3 free for a pawn and gives the knight g3 and f4 later. Rooks going to b1 and c1.
Black has castled short too. Bishop on e7, knights on f6 and d7, rook coming to e8, and the c6 pawn holding d5 in place.
Now count the attacks. There are none. Nothing hits anything. No piece is loose, no king is drafty, no file is open. A student who needs a target inside five moves will manufacture one and lose to it.
White pushes two queenside pawns at Black's three. b4, then a4, then b5. That is the whole plan and it takes a dozen moves of preparation to earn.
The purpose is not to win a pawn and it is not to open a file for the pleasure of an open file. The purpose is to leave Black holding a pawn that no other pawn can ever defend.
There are two outcomes White is happy with. If Black sits still and White gets bxc6 bxc6, Black's c6 pawn is backward on a half open file with no b-pawn or d-pawn left to help it, and White's rooks and knights have somewhere to live for the rest of the game. If Black takes first with ...cxb5 and White recaptures with the a-pawn, Black is left with an isolated pawn on d5, a b7 pawn that needs watching, and an open a-file for whoever gets a rook there first.
The move that ruins all of it is ...c5. Timing is the entire question: too early and Black hands over the centre, too late and there is no centre left to fight for.
The most common student error is pushing b5 on move fourteen because it is the plan and the plan is exciting. Before b5, check three things: your rook is already on b1, the c5 square is covered so the pawn cannot be answered by a comfortable ...c5, and you have somewhere to put a knight after the file opens. If any of those is missing, b5 makes his backward pawn into your weak pawn.
| Plan | Pawn moves | Rooks | Queen and bishops | Knights | Black's usual answer |
|---|---|---|---|---|---|
| Minority attack | b4, a4, b5 | b1 and c1 | Qc2, Bd3, Bg5 (often Bxf6 first) | Nc3 heading a4 and c5, Nge2 | ...a6 and ...Rb8, or ...c5, or trade pieces with ...Ne4 |
| Central break | f3 then e4 | e1 and c1 | Qc2, Bd3, Bf4 or Bg5 | Nge2, Nc3 | ...dxe4 and pressure on d4, or ...Ne4 first |
| Kingside with pieces | f4, sometimes g4 | one on e1, one stays home | Bd3 to b1 with Qc2, Bg5 | Nf3 to e5, Nc3 to e2 to g3 | ...Nf8 to guard h7, ...Ne4 to block, ...c5 to open the middle |
Three plans, and a student who understands them will notice that they share a preparation and diverge late. That is the actual lesson of the structure. You are not committing on move eight. You are improving the position until the position tells you which of the three it wants.
Three answers, roughly in order of how often they work at club level.
Trade pieces. ...Ne4 is the move, and inviting the exchange of dark squared bishops is usually fine. A backward pawn on c6 only matters if there are enough attackers left to sit on it. Four pieces each and the weakness is a fact. Two pieces each and it is a footnote.
Play ...c5 at the right moment. This is the hardest of the three because it requires an honest judgement about whether an isolated pawn is a weakness or a licence for your pieces.
Play somewhere else. ...Ne4, ...Bd6, ...Qd6, ...f5, rook to e6 and across. White's pieces are on the queenside and White's king is not. This is the plan my students enjoy, and it is legitimate, but only after they have proved they can also do the boring one.
There is a fourth, ...b5, freezing the queenside before White is ready. It is respectable and it is dull and it wins games.
Capablanca against Golombek in 1939 is the one everybody shows, and correctly. Black recaptured on d6 with the queen on move 22, and White pushed 23.b5 straight away. Black took, White recaptured with the queen, and from that point the game is about the isolated d5 pawn and the b7 pawn and nothing else. There is no combination in it. That is why it is on the syllabus.
The other one is Pillsbury against Tarrasch at Hastings 1895, and I use it as a warning rather than a model. It opens 1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Bg5 Be7 5.Nf3 Nbd7 6.Rc1 0-0 7.e3 b6 8.cxd5 exd5, so we have the structure with Black's b-pawn already on b6. Then Black resolved it himself with 10...c5 and 11...c4, which ends the Carlsbad argument and hands White a free hand on the other wing. Pillsbury took it: 13.Ne5, 14.f4, and the game finished 52.Qxh7 mate. Tarrasch was not a weak player. He simply decided the structure was uncomfortable and got rid of it, and the bill arrived forty moves later.
Because attacking is a skill and impatience is a habit, and only one of them is chess.
Hal Mireles has a piece on scraping a lathe bed flat (@rustbelt/scraping-a-1943-lathe-bed-flat) where he takes off metal in amounts you cannot see, for days, so that a surface can be true. Nobody watching would call it exciting. It is also the only way that surface ever becomes flat. This structure is the same argument in a different material: you make small improvements that do not look like anything, and then on move thirty two the position is winning and there was no moment where it became winning.
Basirat is 14 and hated this for three weeks. Then she beat a 1750 by pushing b5 on move twenty six, which is late, and she knew it was late, and she waited anyway. She told me afterwards that she nearly played it on move nineteen. That sentence is worth more to me than the game was.
If a student cannot sit inside this structure for twenty moves, he does not have an attacking style. He has an attention span.